StackMap
Subscribe

deja-vu vs memsearch

Indexes the session histories your coding agents already wrote — 17 harnesses, months retroactive — and serves recall over MCP. 84.9% hit@1 on LongMemEval-S, no LLM, no embeddings. One Go binary. — versus — Zilliz's unified memory for coding agents: one Markdown + Milvus store shared across Claude Code, Codex, OpenCode and OpenClaw — hybrid search, plus repeated workflows distilled into skills.

The curated verdict

Unified memory across coding agents, opposite bets: memsearch curates a Markdown+Milvus store going forward; deja mines the raw session logs you already have, no embeddings at all.

deja-vumemsearch
Stars5562.4k
Forks40216
LanguageGoPython
LicenseMITMIT
Last activitytoday5 days ago
Topicsmemory, codingmemory, coding
Curated connections27

deja-vu — the curator's take

The inversion is the insight: every memory tool starts empty and records forward; deja starts FULL from history already on disk, and the no-LLM/no-embedding design means ~1.5ms search, zero keys, zero cost. When NOT: it remembers what your agents did, not curated knowledge — there's no write path for distilled lessons; and its benchmark numbers are self-published, so the standing caution on this shelf applies: benchmark recall on your own corpus.

memsearch — the curator's take

The cross-platform play is the point: a conversation in Claude Code becomes searchable context in Codex, OpenCode and OpenClaw — one memory, four plugins, zero per-agent setup. Memories live in readable Markdown (greppable, versionable) with Milvus doing hybrid search, and the standout feature is procedural: it watches for workflows you repeat and distills them into installable skills, maintained in the background. NOT for single-agent loyalists — if you only run Claude Code, opencode-mem-style plugins are lighter — and it's a Zilliz project: the Milvus dependency is also the funnel; check what 'backed by Milvus' costs you operationally before teams adopt.